Of Shafts with Keyways;

' Cmig & Lund I Study of tk Str i;;^,u :; Of Shafts with Keyways; Mechanical Engineering B.S,....i 9 9 ''''mi- OF

UNIVERSITY OF ILLINOIS LIBRARY Book Volume Ja09-20M

A STUDY OF THE STRENGTH OF SHAFTS WITH KEYWAYS.if?^ (J BY OLLISON CRAIG JAMES CHARLES LUND THESIS FOR THE DEGREE OF BACHELOR OF SCIENCE IN MECHANICAL ENGINEERING IN THE COLLEGE OF ENGINEERING OF THE UNIVERSITY OF ILLINOIS Presented June, 1909

Digitized by the Internet Archive in 2013 http://archive.org/details/studyofstrengthooocrai

1 44877 1 est UNIVERSITY OF ILLINOIS June 1, 1909 THIS IS TO CERTIFY THAT THE THESIS PREPARED UNDER MY SUPERVISION BY OLLISOK GiiAIG JAllES CHARLES LUKD.. ENTITLED A STUDY OF THE STfiENGTH CP SHAFTS WITH _KEYWAYS IS APPROVED BY ME AS FULFILLING THIS PART OF THE REQUIREMENTS FOR THE DEQREE OF BACHELOR OF SGISNGE / Instructor in Charge. APPROVED:-. HEAD OF DEPARTMENT OF MECIIAHIGAL EMINMR.IM.

1. CONTENTS. Page. Introduction.2. General Discussion of Tests and Test Pieces... 5. Description of Apparatus 6. Method of Procedure 8. Formulae and Sample Calculations 10. Drawings and Photos 12. Tables 17. Sample Data 24. Curves 26. Conclusions ^2.

2. Introduction * There has been very little done on the subject of the "Effect of Keyways on the Strength of Shafts'*. Mr. Leidendecker of the Glass of 1908 of the University of Illinois, performed a series of tests on shafts, the tests being in simple torsion alone, and taking no account of the bending effect that comes on the shaft. His work is embodied in his graduating thesis. In the series of tests described in this thesis, combined torsion and bending were applied to the shafts tested. It is of some importance to a designer to know the effect of keyways on the strength of shafts, so that he may allow for the loss of strength thus incurred, in the designing of shafts met with, in machine construction. The tests included tests on shafts with four kinds of keyways; "standard", Extra wide. Extra Deep and keyways to take Woodruff keys. Two sizes of shafts, 1 15/16 inches, and 1 l/4 inches in diameter, were tested in duplicate with above in both combined twisting and bending. Two ratio of bending to twisting were used, one, with a ratio of 6/6 and the other, 6/10, In actual practice shafts are very frequently subjected to a combined torsion and bending which stresses the shaft to a greater degree than if subjected to simple torsion. This may be illustrated by a pulley midway between bearings on a cotint er- shaft, driving a machine below. That is the reason that these tests were made, and that the

I 3, relative amount of twisting moments applied, were varied during the series of tests. According to modem practice, there are two methods commonly used for fastening pulleys to shafts^by set screws and by keys. The set screw is convenient where the power to be transmitted is small but otherwise the use of keys is far more satisfactory. There are several ways in which keys are used. In some cases they are put in merely for the purpose of keeping the pulley from rotating on the shaft, thus driving by shearing action alone. In other cases the keys are used to prevent pulleys from sliding along the shaft as well as for the transmission of power. In the latter case they are machined all over so that they fit on all sides. This necessitates a taper and they must be driven in tightly, often eauses a severe bursting strain in the hufc of the pulley. But the fitting of taper keys is expensive so they are not used as much as the square keys, which are made to bear tightly on tv/o sides with a slight clearance on the top and bottom. Where shafts are made for sliding bearings as in the case of Drilling Machine Spindles, the depth of the keyway is generally made greater than in cases in which the pulley does not slide and when there is a heavy twisting moment to be transmitted, two keys are sometimes placed at right angles to each on the shaft. During later years the V/oodruff key has come into prominence and for many purposes is eminently satisfactory. These keys shown on page 12 Fig. 4. But it has some disadvantages, one of 1^ I

4. which is the great amount of trouble required to remove a key after it has once been fitted into place.

General Discussion of Tests and T_est Pieces, In the series of tests performed, a standard was assumed. This was done, because different Manufacturers use different standards and it was thought best to assume a keyway, fairly well averaging the dimensions of the different standards. The keyway assumed as a standard was based on the general formula; l/4 d in breadth, by 1/8 d in depth, in which "d" represents the diameter of the shaft. This would provide for a square key, the dimensions, of which, would be one-fourth of the diameter of the shaft. The sizes of the Woodruff keyways were selected by computing the twisting force at the s\irface of the shaft for thie maximum loads, and selecting Woodruff keys which were strong enough in shear to resist this force. The keys were selected by dividing the twisting force at the surface of the shaft by the allowable unit shearing stress of the material of which Woodruff keys are made, and choosing keys, v^hose cross-sectional area at the surface of the shafts equaled those quotients. The general appearance of Woodruff keyways is shown in the drawings page 12 Fig. 4 The different variations of tests were as follows 1. Shafts having a diameter of 1 15/16 inche. keyways in horizontal plane, ratio of twisting to bending 10 to 6. 2. Shafts having a diameter of 1 15/16 inch, keyways in horizontal plane, ratio of twisting to bonding 6 to 6.

6. 3. Shafts having a diameter of 1 15/16 inch, keyways in vertical plane, ratio of twisting to bending 10 to 6, 4. Shafts having a diameter of 1 15/16 inch, keyways in vertical plane, ratio of twisting to bending 6 to 6, 5. Shafts having a diameter of 1 l/4 inch, keyways in vertical plane, ratio of twisting to bending 10 to 6. 6. Shafts having a diameter of 1 1/4 inch, keyways in vertical plane, ratio of twisting to bending 6 to 6, All variations of keyways used in each of the six series. All test ii made in duplicate. All shafts 30 inches in length. Description of App aratu s. The machine used in all the combined stress tests was the Philadelphia, 100,000 pounds capacity, in the Laboratory of Applied Mechanics at the University of Illinois. A special apparatus designed and built in the laboratory was attached to this machine which made it possible to secure combined stress. Such an apparatus is shown on page 15,and diagramatically on page 14. The shaft with keyways in the ends, is keyed to the rocker arm "a" as shown. The apparatus is then rested in bearings "b" which are supported by steel ball bearings in a race in the block "c". These blocks in turn are placed on plates "d" which are supported by two small I beams which rest on the weighing head of the testing machine. The blocks "c" with plates **d" are movable so as to vary the bending moment for the different ratios of

bending to twisting. The twisting moment is varied by shifting the tension! rods **e" along the rocker arms "a". These arms have spherical recesses *m" which give different arm lengths into which the spherical pointed projections on the tension rods fit. All other parts are rigidly fixed. Over the ends of the rocker arms "a" are hung the tension rods "e" Y/hich at their lower ends support an I Beam "f** fastened to the movable head "g" of the machine. In measuring the twist deflections indicators "h" were arrange[l as shown in the photograph Pig. 10 These indicators were specially designed for this particular series of tests performed in this thesis. They consisted of three long arms clamped on to the shaft and equally spaced along the shaft. One arm v/as clamped at the center, and the other two at equal distances as to span with the center arm, the part of the shaft including the keyway and a solid part, respectively. At the extreme end of the last two arms mentioned were placed scales glued ^to mirrors. The center arm held pointers, that these scales passed under as the shaft twisted* Readings were taken fron the scale in fiftieths of an inch and by reading the instrument when the pointer covered its own reflection, errors due to paralla? were avoided. The approximate length of the indicator arms necessary, was computed from the degrees twist, that was expected; and the distance between arms which was 5.5 inches. When the testing of the 1 l/4 inch shafts was begun, it was noticed that the machine was too inaccurate for close work, the

8» balance beam being too sluggish for the,* applied load, A needle beam was attached to the machine, immediately above the regular beam. This proved to be very satisfactory and the sensitiveness of the machine was such that the load on the specimen could be determined to within 10 pound load. Load was applied in increments of 1000 pounds, for the 1 15/l(i inch shaft and of 300 pounds for the 11/4 inch shafts. After readings were taken the load was taken off, until the initial load was reached. This gave a means of determining whether there had occurred any set or deformation in the shaft. This was continued until a set was obtained. Prom the readings so taken, we were able to plot curves showing relation between the stress applied and the set and the deformation suffered by the shaft, for both keyway and solid side of each shaft. All data taken during the test were recorded in a log book which is on file at the Research Office of the Laboratory of Appliedl Mechanics. A sample of notes taken is shown on page 24, also the calculated value as computed and used in working up results. Method of Procedure. 1i\e test pieces used 30 inches long and of two different sizes in diameter, 1 15/16 inches and 1 1/4 inches. In the series of tests, the ratio of bending to twisting was 6 to 10. At first, some of these shafts were tested with the keyv/ay in a horizontal plane, but owing to the fact that the neutral axis was in the same

j 9. plane, the keyway was not subjected to the maximum bending stress so that the later tests were all made with the keyway s in the vertical plane. When the apparatus was rigged up a load of 500 pounds was applied and was regarded as the initial or zero reading. The initial reading on the keyway side and on the solid side were noted, and the deflectometer "M** placed under the center of the shaft, its dial hand pointing to zero. The deflectometer readings gave approw imately the amount of bending. The shafts were tested in such a manner as to get the points i of set. In order to do this, an initial load was taken, and each time, after the load increment was added, the machine was reversed and the load brought to the initial load. In this manner a series of points were obtained. As long as the readings of twist were the same at the initial load, there was no set in the shaft, but when the readings began to increase at the initial load, the shaft was beginning to take a set. The point at which a set first took place is called " The point of first set", and this point would correspond to a given load. The load corresponding to the point of first set in the load, up to which, the shaft may be stressed, without causing it to take a set, that is it locates approximately the elastic limit. The material in the different shafts used, was not uniform, but that fact was judged to not effect, seriously, the results, as the strength of one shaft was not compared to that of another but rather the strength of a portion of a shaft, containing a

10. keyway, was compared to that of a portion of the same shaft, which did not contain a keyway. Preliminary calculations were reduced to constants. Thus, in mailing calculations for the degrees twist; knowing the length of the indicator arm, the distance the indicator arms were apart on the shafts, and the length of the spaces in inches, on the scale, a constant was found, which, when multiplied by the scale readings, gave the degrees twist per inch of shaft. Similarly a constant was found, knowing the length of the twisting arm and the size of the shaft, which, when mulitiplied by the load, gave the unit stres in the metal. Formulae and Sample Calculations. ( a ) = P-T = SJ. which 2 c M = Twisting moment in pound-inches. P = Load on scales. T = Length of twisting arm in inches. S = Unit stress in metal. J = Polar moment of inertia of shaft, c = Distance from neutral axis to extreme fiber or, in this case, the radius of the shaft. Substituting in the preceding formula for case of shaft 1 15/16 inches in diameter, ratio of bending to twisting 6 to 10. ^ ^ ^ - ^ ^ T0688 * S - 0.5 P

11. ( b ) Derivation of constants for degrees twist per inch. - L -0- = \ -, m which Li -9- = Angle of twist in radians. A = Scale readings in spaces. L = Distance from center of shaft to scale. Substituting for case of shaft 1 15/16 inches in diameter. L = 37.375 inches. = radians. 37!375 X 5.5 ' 5.5 indicates the distance apart in inches the indicators were set on the shafts. 50 divisions on scale = 1 inch. in degrees per inch = ^ of length, of shaft = 005575 A.

12. Pig. 4. 15 Diam. of Shaft, 1 inches. Woodruff Keyway No. S, 5/l6 in. x 5/8 in.

13 Fig. 7. Diam. Shaft, 1 ^ inches. Extra V7ide Keyway, in* x in. 8 Pig. 8. Diam. Shaft, 1 t inches. 1 7 Woodruff Keyway No. 15, ^ inch, x ^ inches.

V

14. Pig, 9. Diagramatic Sketch of Testing Apparatus Combined Torsion and Bending.

Fig, IC, Apparatus for Testing Shafts in Combined Torsion and Bending,

16, Figa 11* Shafts after having been Tested in Combined Torsion mid Bending*

17 Table No. 1. Diameter of Shaft = 1 i& inches. Horizontal Plane. Ratio = rr. 16 10 Dimensions of Keyway i x i 2 4 inches. Standard Extra Wide Extra Deep V/oodruff X i 4 1 x3 2 8 2 No. S. Stress at point of 1 first Set lb. per sq. 2 in. Keyway Side. 29750 26250 26250 26250 26750 33250 22750 22750 Stress at point of first Set lb. in. Solid Side. 1 per sq. 2 36750 29750 29750 33250 29750 36750 32375 32375 Loss of Strength 1 19 12 10.1 29.3 Percent. 2 12 21 9.55 29.3 Average loss of Strength, 15.5 16.5 9.8 29.3 Percent.

18. Table No. 2. 15 6 Diameter of Shaft = 1 TB" inches. Horizontal Plane. Ratio = g. Standard Extra Wide Extra Deep Woodruff Dimensions of Keywayinches. 1 X 2 1 4 X 1 1 X 3 2 8 2 No. S. Stress at point of 1 first Set lb. per sq. 2 in. Keyway Side. 15750 32550 30 450 13650 24150 Stress at point of 1 first Set lb. per sq. 2 in. Solid Side. 17850 32550 32550 15750 30450 Loss of Strength 1 11.8 6.5 13.3 Percent. 2 20.7 Average loss of Strength, 5.6 6.5 17. Percent.

19. 1 laole wo. L>iainst.er oi onait J. inches. Vertical Plane. Ratio = ot* anclara Extra Wide Extra Deep V/oodruff Dimensions of Key ways inches* 1 1 1 X 1 1 3 2 8 2 No. S. Stress at point of first Set lb. per sq. m. xveyway oicie. Stress at point of first Set lb. per sq. m. Solid Side. 1 22750 15750 f57no 22750 2 19250 26750 22750 1 29750 26750 12250 2O750 2 26750 33250 2P750 JjOSS 01 otrengtn 40 28.7 25.6 Percent. 26.7 23.6 Average loss of olreng ^n 40 27.7 23.6 Percent.

20. Table No. 4. 15 6 Diameter of Shaft = 1 "Jg inches. Vertical Plane. Ratio = g Standard Extra Wide Extra Deep Woodruff Dimensions of P^eyways inches. 2 4 3 8 2 No. S Stress at point of first Set lb. per sq. in. Keyway Side. 15750 26250 19P50 22050 19950 1P950 26250 28350 Stress at point of first Set lb. per sq. in. Solid Side. 17850 32550 24150 28350 28350 30450 32550 Loss of Strength 1 11.8 39.8 29.6 13.7 Percent. 2 19.4 8.7 29.6 12.9 Average loss of Strength 15.6 24.2 29.6 13.3 Percent.

21. Table No. 5. 1. Diameter of Shaft = 1 t ir^ches. Vertical Plane. Ratio = Standard Extra Wide Extra Deep Woodruff Dimensions of Keyway inches. 5x5 16 32 8 5x5 32 5 X 16 15 2 N6. 15 64 Stress at point of 1 first Set lb. per sq, 2 in. Keyway Side. 30 6 22 26539 30622 30622 26539 34705 34705 38787 Stress at point of 1 first Set lb. per sq. 2 in. Solid Side. 34705 306 22 34705 34705 34705 34705 34705 Loss of Strength 1 11.8 11.8 13.3 10.6 Percent. 2 13.3 11.8 10.5 Average loss of Strength 12.5 11.8 11.8 5.3 Percent.

22. Table No. 6. Diameter of Shaft = 1 ^ inches. Vertical Plane. Ratio =. Standard Extra Wide Kxtra Deep Woodruff Dimensions of Keyway inches. 16 32 5 X 5 5 ^ 15 8 32 16 64 2 No. 15 Stress at point of first Set lb. per sq. in. Keyway Side. 28175 28175 23725 20825 28175 28175 30625 28175 Stress at point of first Set lb. per sq. in. Solid Side. 30625 3307 5 37975 35525 37975 35525 37975 33075 Loss of Strength 1 8 38.7 47.3 19.4 Percent. 2 14.8 41.4 22.5 14.8 Average loss of Strength 11.4 40 34.9 17.1 Percent.

23. Table NO. 7. General Averages of Loss of Strength. ( in percent ) Standard Keyway 14.15 Double Width " 29. 3 ^ Stranded Depth " 26. Woodruff * 14.8 Table No. 8. Loss of Strength in Percent. Standard Key ways. Combined Stresses. Simple Torsion. Size of Shaft Ratio of Ratio of 6 to 10 6 to 6 ^5 25.1 15.6 2. 1 Yq inches. ( Leidendecker ) Size of Shaft 1 i inches. 4 12.5 11.4 4.5

24. Sample Data Shaft No. 6 Load A B C Unit Stress Twist Degrees per lb* lb. per sq.iq. incvi > A B 500 30 13 1750 1000 32 14 1 3500.01112.0055 500 30 13 1750 2000 37 18 3.5 7000.039.0279 500 30 13 1750 3000 40 21 6 10500.0557.0446 500 30 13 1750 4000 45 24 8.5 14000.0835.0613 500 30 13 1750 5000 49 27 11 17500.106.0725 500 30 13 1750 6000 52 30 14 21000.1225.0948 500 30 13 1750 7000 58 33 17 24500.156.1112 500 31 13-1 1750.0055 8000 72 36 22 28000.234.128 500 40 13 1 1750.0505 9000 160 40 50 31500.725.1505 500 124 13 25 1750.524 10000 45 35000.1782 500 16 1750.0167

Sample Data. ( continued ) Duplicate Extra Wide Length 30 inches Diameter 1 15/16 inches Size of Keyway 1 in. x l/4 in. Twisting Arm 10 inches Bending Arm 6 inches. Ratio A. - Left hand scale - ( Keyway ) B. - Right hand scale -( Solid shaft ) C. - Deflectometer - ( Center of shaft ) KeywEy set in Horizontal Plame.

EUGENE DIETZCEN CO.. CHICAGO.

' 32. Conclusions,! The first set, in almost every case, took place under the load corresponding to the yield point. 2. The shafts were materially weakened by having the keyways in the vertical plane rather than in the horizontal plane. When in the vertical position the keyway was in that part of the shaft subjected to the greatest stress due to bending, but when it was in the horizontal position it was practically in the neutral axis. 3. There was considerable loss in stiffness as indicated in the following table. The table gives the loss of stiffness in percent. Ratio of Twisting 1 inch. Diam. 1^ inch. Diam. to Bending 10 to 6 13.3 9.5 6 to 6 20.4 14.9 4. In general the keyway s had a weakening effect in the following order J- Extra wide 29 percent. Extra deep 26 " " Woodruff 14.8 " " Standard 14.15 The extra wide keyv;ays had the greatest weakening effect, due th the fact that the most metal was removed from the part of

33. the shaft that was subjected to the greatest torsional stress. Contrary to general opinion, the V/oodruff keyways, next to the standard keyways, showed the least weakening effect, because in cutting a Woodruff keyway, more metal is taken from nearer the axis of the shaft, and not so much from the outer fibres, which are undei' the greatest torsional stress. 5. In comparing the results of this series of tests with those obtained by Mr. P. E. Leidendecker, of the Glass of 1908 of the University of Illinois, in a series of tests in torsion alone, it is plainly shown that a keyway has a greater weakening effect if the shaft is subjected to both torsion and bending. 6. It would be well, when selecting a shaft, to make allowance for the weakening effect of keyways. This could be done by assuming a working stress of 15 percent or 20 percent lower than that used in usual calculations.